转自：前瞻产业研究院 智能营销、教学辅助、3D建模以及智能驾驶等应用场景是生产生活中的重要领域，也是目前多模态大 模型可以切入并且精准赋能的领域。根据赛迪四川研究数据显示，2024年智能营销占中国人工智能多模 态大模型20强企业模型场景的9.5%，教学辅助、3D建模和智能驾驶均占4.8%左右。 行业主要上市公司：阿里巴巴(09988.HK，BABA.US);百度(09888.HK，BIDU.US);腾讯(00700.HK， TCEHY);科大讯飞(002230.SZ);万兴科技(300624.SZ);三六零(601360.SH);昆仑万维(300418.SZ);云从科技 (688327.SH);拓尔思(300229.SZ)等 本文核心数据：应用场景比重; 多模态大模型生成生活相关场景 多模态大模型助力智能营销优化策略 智能营销行业利用人工智能、大数据、机器学习和多模态技术，通过自动化、个性化的方式优化广告投 放、客户关系管理和内容营销。智能营销不仅帮助品牌实现更高效的客户触达，还能够动态调整营销策 略，提升用户体验，推动品牌增长。 智能营销是应用人工智能技术，对数字营销的全链路进行智能化升级的新型营销方式。智 ...

云从科技自主研发的"从容多模态大模型"在Open Compass评测中以65.5分位列全球前三，超越谷歌 Gemini1.5Pro等模型，并在跨模态跟踪、3D人脸识别等细分领域10次刷新世界纪录。基于这一技术优 势，公司与华为昇腾联合推出的智用一体机解决方案，已在天津港（600717）智慧物流调度、国网山东 能源管理等多个标杆项目中落地，助力企业运营效率提升超20%。(36氪) ...

Core Insights - Hong Kong has established the Von Neumann Institute to integrate embodied intelligence, generative AI, and advanced supercomputing technologies, aiming to promote interdisciplinary collaboration and commercialize research outcomes [1][2] - The institute focuses on five key AI areas: multimodal AI systems, enhancing AI reasoning capabilities, robotics intelligence, AI-driven 3D understanding, and healthcare service reform through large models [2][3] - The institute aims to build a talent pipeline in AI through educational initiatives, targeting over 100 PhD students and engaging with local schools to foster innovation [3] Group 1 - The Von Neumann Institute is the first "full-chain practical" AI research institute in the Greater Bay Area, bridging basic research and industrial application [2] - The institute's approach includes establishing specialized laboratories and joint industry-university collaborations to accelerate the transition from theoretical research to product development [2][3] - The leadership of Jia Jiaya, who has experience as a scientist, entrepreneur, and educator, positions the institute to become a model of "top-tier research and grounded industry" [3] Group 2 - Jia Jiaya emphasizes Hong Kong's role as an "innovation brain" due to its international capital flow, top-tier research resources, and global talent hub, while Shenzhen acts as the "industrial driver" [4] - The integration of large models and visual sensor hardware by Simo Technology showcases a demand-driven approach to innovation, with applications in major factories like Tesla and BYD [4][5] - The collaboration between Hong Kong and Shenzhen has created a rapid coordination mechanism, allowing for quick transitions from research to production [5]

中新社上海5月8日电 (记者郑莹莹)上海徐汇区"AI+文旅生态成长计划"8日在模速空间内启动。参与活动 的业界人士认为，对于拥抱人工智能技术，文旅行业更具包容度。 "工业等领域的应用场景需要非常高的准确率，但文旅场景对于这类新科技的包容度是比较高的。比 如，机器人表演有时还会摔跤，对此大家其实是能包容的。"上海魂伴科技有限责任公司(简称：魂伴科 技)合伙人金成思说。他认为，文旅场景有望率先实现人工智能应用落地。 魂伴科技在2025年4月举办的2025上海龙华庙会上展示人形机器人应用。 中新社记者郑莹莹摄 魂伴科技在今年4月举办的2025上海龙华庙会上"秀"了一把机器人，吸引了沪上众多市民游客围观。这 对金成思触动很大："当时机器人的表演其实并没有往日视频里酷炫，但市民游客仍觉得比在网络视频 里看到的更好、更真实，现场有些老年市民看到现实版人形机器人后，还期待它未来能帮忙养老。" 这让他思考，也许更重要的是让更多市民有机会了解、接触机器人。"我们要让机器人产品从实验室里 走到广场上，了解市民的需求，哪怕让机器人出洋相。如此，我们才能知道我们差的是什么。" 上海稀宇科技有限公司的公共事务副总裁严奕骏也看好文旅领域 ...

投资建议。 人形机器人高速发展，具身智能是驱动商业化落地的核心因素。机器人智能水平以及实时控制 性能提升将驱动感知性能、算力、通信效率等需求增长，端侧传感、驱控及通信芯片将充分受益。具身智 能落地打开人形机器人成长空间，未来应用前景广阔，带动整机厂商业绩上行。 报告导读： 具身智能是人形机器人商业化落地核心，多模态、强化学习加速智能进化，感 知传感迭代革新， EtherCAT 赋能高速通信，端侧算力持续升级。 本文摘自：2025年5月8日发布的 从"能动"到"灵动"，机器人智能化步入新篇章 舒 迪 ，资格证书编号： S0880521070002 更多国泰海通研究和服务 亦可联系对口销售获取 重要提醒 本订阅号所载内容仅面向国泰海通证券研究服务签约客户。因本资料暂时无法设置访问限制，根据《证 券期货投资者适当性管理办法》的要求，若您并非国泰海通证券研究服务签约客户，为保证服务质量、 控制投资风险，还请取消关注，请勿订阅、接收或使用本订阅号中的任何信息。我们对由此给您造成的 不便表示诚挚歉意，非常感谢您的理解与配合！如有任何疑问，敬请按照文末联系方式与我们联系。 法律声明 市 场空间超万亿，实现具身智能是商业化落 ...

Group 1 - The core viewpoint is that embodied intelligence is the key to the commercialization of humanoid robots, with a market space exceeding one trillion yuan, and the intelligent level of humanoid robots in China is expected to evolve significantly by 2045 [1] - Humanoid robots possess human-like perception, body structure, and movement, making them highly adaptable to human society, with potential applications in manufacturing, social services, and hazardous operations [1] - The market scale for humanoid robots is currently below ten billion yuan, but as intelligent levels progress towards embodied intelligence, the market is expected to expand significantly [1] Group 2 - Multi-modal large models and reinforcement learning are enhancing operational control performance, with significant advancements in communication and computing power to support real-time control [2] - Major companies like NVIDIA and Tesla are integrating multi-modal perception to improve robot interaction and decision-making accuracy, while the development of embodied reasoning models is expected to enhance performance in complex environments [2] - The adoption of pure visual solutions and advanced sensors is anticipated to lower hardware costs and improve perception sensitivity, with EtherCAT emerging as a mainstream communication protocol due to its high real-time performance [2]

Core Insights - The multi-modal large model industry focuses on deep learning models capable of processing, understanding, and generating various types of data, including text, images, audio, and video, enabling complex and intelligent tasks [1] - The industry has a wide application potential across various sectors such as natural language processing, image recognition, speech recognition, intelligent driving, and medical imaging diagnosis [1] Industry Overview - The multi-modal large model industry chain is complex, encompassing hardware facilities, software development, and various model types, including CLIP, BLIP, and LLaMA, among others [1] - The industry is divided into three layers: the foundational layer (hardware and basic software), the model layer (various types of multi-modal large models), and the application layer (industry-specific applications) [1] Cost Structure - The training costs for mainstream domestic large models range from tens of millions to hundreds of millions of dollars, with major companies like Baidu, Alibaba, and Tencent investing over $200 million [3][5] - Startups like Kimi and DeepSeek have managed to reduce training costs to between $30 million and $60 million through technological optimizations [3] - Cloud hosting costs are significantly influenced by model scale, with major companies leveraging their own cloud platforms to reduce costs [3] Development History - The global large model industry has evolved through several phases: early exploration (1956-2005), rapid growth (2006-2019), the rise of large models (2020-2022), and the current phase of widespread application starting in 2023 [6] Computational Demand - The demand for computational power in AI is increasing, with larger models requiring exponentially more computational resources; for instance, the GPT-3 model requires 3640 PF-days of computation and at least 10,000 GPUs [9] - As model parameters increase, the computational investment needed grows significantly, influenced by model architecture, optimization efficiency, and hardware capabilities [9]

转自：前瞻产业研究院 行业主要公司：阿里巴巴(09988.HK，BABA.US);百度(09888.HK，BIDU.US);腾讯(00700.HK， TCEHY);科大讯飞(002230.SZ);三六零(601360.SH);云从科技(688327.SH)等 本文核心数据：多模态大模型代表企业融资规模;多模态大模型代表企业投资规模 2025年开始投融资呈爆发式增长 截至2025年4月，多模态大模型投融事件数量接近50件，其中国2021年投融资金额出现了高峰，达19.1 亿元，尽管当年投资事件数量为5件。2024年开始新一轮的投资周期，共有11件投资事件，金额达5.16 亿元。2025年前4个月，共有17件投资事件，金额为16亿元，后续多模态大模型题材的投资将呈现爆发 式增长。 企业能获得多轮投资 根据IT桔子显示，多模态大模型行业2025年开始投融资恢复热度。主要的融资事件如下： | 时间 | 254 | 地区 | 在不同分 | 金额 | 融资金额 | 投资方 | | --- | --- | --- | --- | --- | --- | --- | | 2025/4/9 | 爱芯元智 | 宁波市 | 人工智 ...

Core Viewpoint - The audio industry in China is experiencing significant growth driven by technological advancements, particularly in AI and multimodal models, enhancing content creation and user experience. The market size is projected to reach 28.7 billion yuan in 2024, reflecting a year-on-year growth of 14.80% [2][9]. Industry Overview - Audio refers to sound signals perceptible to the human ear, typically ranging from 20 Hz to 20,000 Hz. It can be classified into analog and digital audio based on the signal format [2]. Industry Development History - The Chinese audio industry has evolved through four main stages: - **Incubation Period (1996-2005)**: Initiated with Guangdong Pearl River Economic Broadcasting's real-time online broadcasting in 1996 and the introduction of podcasts by Apple in 2005 [4][5]. - **Exploration Period (2006-2015)**: Marked by the launch of early Chinese audiobook websites and regulatory frameworks, including the establishment of Douban FM and the founding of Ximalaya [4][5]. - **Expansion Period (2016-2019)**: Characterized by the introduction of live streaming features by major platforms like Ximalaya and Lizhi, intensifying competition [4][5]. - **Maturity Period (2020-Present)**: Notable events include Lizhi's IPO in the U.S. in 2020 and advancements in AI applications in audio systems, indicating a shift towards comprehensive AI integration in the industry [4][5]. Industry Value Chain - The audio industry value chain consists of: - **Upstream**: Content creation (music, audiobooks, podcasts), raw materials (metals, plastics), and components (resistors, capacitors, microphones) [6]. - **Midstream**: Audio platforms that facilitate content distribution [6]. - **Downstream**: Various listening channels including smartphones, smart speakers, and wearable devices, along with the end-users [6]. Related Companies - Key listed companies in the audio sector include Tencent Music (01698), NetEase Cloud Music (09899), and Edifier (002351), among others [2].

Group 1 - The report by Zhong An Technology and Zhong An Financial Technology Research Institute explores the application of large language models (LLMs) in the financial and insurance industries, concluding that LLMs present new opportunities but face challenges in implementation that require multi-party collaboration [1] - The development of large model technology is diversifying globally, with vertical models emerging to provide tailored industry solutions. China has made progress in computing autonomy and data optimization, leading to a trend of functional differentiation and specialization in its ecosystem [1][24] - New technologies are driving down the costs of training, operation, and inference for large models, prompting a restructuring of processes in the financial industry. Financial enterprises need to balance acquisition, inference, and operational costs while selecting appropriate deployment models and roles [1][12] Group 2 - Domestic models like DeepSeek and Tongyi Qianwen have achieved breakthroughs in cost control and inference performance, providing better technical options for insurance institutions while ensuring data security and compliance [1][15] - Insurance institutions are accelerating the integration of large models, focusing on internal efficiency improvements across the entire insurance business chain and back-office management. Caution is advised during pilot applications to address data security and AI hallucination issues [1][16] - The value of data elements is becoming more prominent, with the financial and insurance industries building high-quality datasets through horizontal, vertical, and government-enterprise collaboration mechanisms to promote intelligent transformation [1][19] Group 3 - The application of large language models in the financial and insurance sectors is transitioning from pilot exploration to systematic integration, with initial deployments focusing on low-risk, low-intervention auxiliary business scenarios such as intelligent customer service and smart claims [6][7] - The introduction of large language models is not only enhancing process efficiency but also driving a deep transformation in information processing paradigms and decision-making logic within the industry [8][9] - The rise of large language models is reshaping the operational philosophies, business logic, and value creation models of financial institutions, leading to trends such as precision financial services and cross-industry ecological collaboration [9][10] Group 4 - The evolution of large model technology is characterized by a shift from purely algorithmic breakthroughs to the construction of systemic capabilities that integrate model deployment, business processes, and system interfaces [29][30] - The deployment capabilities of large models are transitioning from "usable" to "adaptable," with future competition likely focusing on building flexible deployment mechanisms across architectures and scenarios [31] - The emergence of vertical large models is addressing the specific needs of industries like finance and healthcare, enhancing precision and efficiency in tasks such as risk assessment and compliance checks [40][41]